Bulk photovoltaic and photoconductivity effects in two-dimensional ferroelectric CuInP₂S₆ based heterojunctions

Two-dimensional materials are expected to bring about a significant revolution in the fields of optoelectronics, microelectronics, and energy devices,^1–5 owing to their extremely thin thickness and diverse physical and chemical properties. They have the potential to enable high-density, low-power consumption, and flexibility,^6–8 which makes them to be highly desirable materials for use in future devices, especially in the areas of optoelectronics and microelectronics.^9–11 These advancements could lead to breakthroughs in Moore's Law, paving the way for possibilities in technological innovation. What is particularly noteworthy is the construction of two-dimensional heterojunctions, which offers a tremendous expansion in the modulation degrees of freedom available for two-dimensional materials.^12–14 It has led to the creation of a wide range of advanced microelectronics and optoelectronic devices.^12,15,16 Two-dimensional heterojunctions are able to facilitate p–n interface modulation, bandgap modulation, and carrier concentration modulation, all of which contributes to the excellent photodetection performance in photodetectors that incorporate two-dimensional heterojunctions.^15,17 The photovoltaic and photoconductivity effects are essential physical mechanisms used in modulating and achieving excellent photodetection performance in two-dimensional heterojunction-based photodetectors.^18–20 Extensive research has been conducted on these effects. However, the bulk photovoltaic effect, which is a type of nonlinear optical process that occurs in the photovoltaic effect, has not received enough attention. Since it is only present in materials with broken inversion symmetry, leading to significant ferroelectric polarization.²⁰ The bulk photovoltaic effect has immense potential in applications of self-powered optoelectronics, microelectronics, and energy conversion devices. It would provide opportunities for developing future devices to conduct the research on bulk photovoltaic effects in two-dimensional ferroelectric heterojunctions.

In this report, a two-dimensional ferroelectric heterojunction is constructed with multi-layered CIPS and MoS₂ nanoflakes, and the photoconductivity effect and bulk photovoltaic effect are investigated in detail. Although the wide bandgap (2.62 eV) of CIPS suppresses the photodetection performance,²¹ the ferroelectric polarization promotes the bulk photovoltaic effect in the two-dimensional ferroelectric heterojunction. Furthermore, the effect of pyroelectricity is excluded in the short-circuit photocurrent (I_sc) of the two-dimensional ferroelectric heterojunction, proving the polarization induced bulk photovoltaic effect. The MoS₂ is a promising two-dimensional material for high-performance visible photodetectors, whose stable phase is the centrosymmetric 2H phase, which limits its ferroelectricity. In all, the photoelectric effect in the MoS₂ layer contributes to the photoconductivity effect of the heterojunction, while the room-temperature polar ordering in CIPS contributes to the bulk photovoltaic effect. The heterojunction exhibits high D^* of about 1.89 × 10⁹ Jones, when the optical power intensity is 4.71 mW/cm². Moreover, the short-circuit photocurrent density is high, reaching about 1.23 mA/cm² when the optical power intensity is 0.35 W/cm². This work highlights the potential application of two-dimensional ferroelectric materials in multifunction devices with detection and energy conversion capabilities.

The heterojunction was fabricated using the dry-transfer technique. The Au electrodes were prepared on a Si substrate with 300 nm SiO₂ coated on the top by using the small ion sputtering instrument (SBC-12). The thicknesses of CIPS and MoS₂ crystals were repeatedly reduced via mechanical exfoliation using a blue tape. The vdW heterojunction were made by the two-dimensional transfer platform (E1-T). The resulting CIPS flake on the tape was transferred onto one side of the Au electrode. This was done with the help of a polydimethylsiloxane (PDMS) film on a two-dimensional transfer platform. The MoS₂ flake was then transferred over the CIPS flake and onto the other side of the Au electrode using the same method. The PDMS should be heated to 80 °C for 5–10 min until the flakes are thoroughly transferred to the substrate. The topography of fabricated heterojunctions was obtained by an optical microscope (DMM-900C). The TEM and SAED images were obtained by the transmission electron microscope (JEM-2100). The current–voltage (I_ds–V_ds) curves were measured through an Agilent B1500A semiconductor device analyzer with a probe station (EB8). The light sources were provided by Changchun New Industries Optoelectronics Tech Co.

Figure 1(a) illustrates the structure of the two-dimensional photovoltaic heterojunction with a sandwiched structure. The crystal structure of CIPS is defined by the sulfur framework in which the Cu, In, and paired P–P atoms fill the octahedral voids.^22,23 Bulk crystals are composed of vertically stacked and weakly interacting layers linked by vdW interactions.^24,25 The off-center shift of the Cu cations leads to stable out-of-plane ferroelectricity, and the electrically switchable polarization in CIPS has been demonstrated down to the thickness of 4 nm at room temperature.^21,22 Therefore, ferroelectric CIPS with controllable thickness are an ideal candidate for enhanced photocurrent generation via the bulk photovoltaic effect. The inset of Fig. 1(a) shows the characteristic I–V curve of the bulk photovoltaic effect, and the black and red lines represent the I–V curves at the dark and illuminate conditions, respectively. Figure 1(b) shows the optical image of the CIPS/MoS₂ heterojunction, in which the region of top MoS₂, CIPS, and bottom Au electrode are marked with purple, green, and dark dashed lines, respectively. The overlap area of MoS₂, CIPS, and Au electrode is circled by the red dashed line. As shown in Fig. 1(c), a capacitor-like structure is prepared in which the dielectric layer or polar material is sandwiched by the top and bottom electrodes, MoS₂ flakes, and metal gold, respectively. Figures 2(a) and 2(b) present the high-resolution transmission electron microscope (TEM) images of the CIPS and MoS₂ nanoflakes, respectively. The insets illustrate the selected area electron diffraction (SAED) patterns with multiple sets of symmetry patterns and a perfect hexagonal structure.^3,13

Figure 3(a) shows the I_ds–V_ds characteristics of the heterojunction under 405 nm laser with different power densities as bias within the range of ±3 V. It is obviously that the heterojunction exhibits improved photocurrent responses with the optical power intensity rising from 0 to 0.35 W/cm². Under the optical power intensity of 0–0.35 W/cm², the current is 9.38 × 10⁻⁹–1.22 × 10⁻⁷ A at the bias of 3 V. The asymmetry and nonlinear shapes of the I_ds–V_ds curves demonstrate the formation of vdW heterojunctions. When the laser is turned on, the photo-induced electron and hole pairs will be generated and separated by the built-in ferroelectric field,^26,27 thus forming a stable photocurrent. Figure 3(b) shows the J–V curves near zero with light illumination, and short-circuit current density (J_sc) is the photocurrent density from the I_sc divided by the capacity area. At the optical power intensity of 0.35 W/cm², the J_sc is about 1.23 mA/cm².

To further analyze the mechanism, Figs. 4(a) and 4(b) illustrate the value and trend of J_sc and open-circuit voltage (V_oc) of the heterojunction with different power densities. When the heterojunction is irradiated with laser, a non-zero positive J_sc and negative V_oc could be observed without external bias. The zero-bias photocurrent is regarded as the characteristic feature of the bulk photovoltaic effect.²⁸ In Fig. 4(b), the maximum value of V_oc is estimated to be −0.085 V at the optical power intensity for 0.16 W/cm² that is about 301 K, and then the V_oc begins to decrease as the optical power intensity increases. It can be explained for that CIPS is a room-temperature ferroelectric with the Curie temperature at 315 K, around which the ferroelectricity of CIPS decreases.²³ The photovoltaic effect in the CIPS/MoS₂ heterojunction confirms the potential application of two-dimensional ferroelectric materials in self-powered multifunction devices.

When the laser irradiates the CIPS/MoS₂ heterojunction, the phonons in the semiconductor are released, resulting in local temperature increase in the lattice.²⁹ Therefore, we measured the J_ds–V_ds curves at different temperatures at dark condition within the bias range of ±3 V. Figure 5(a) shows that the current response is gradually upward with the increase in temperature, and Fig. 5(b) displays the J–V curves near zero. The inset shows that the J_sc of the device gradually decreases with the increase in temperature, which is contrary to the trend caused by light, indicating that heat has an inhibitory effect on J_sc. It is not difficult to find that lower temperature did not greatly affect J_sc while the higher temperature did. As seen in Fig. 4(a), when the optical power intensity is low, the J_sc increases rapidly with the increase in the optical power intensity, and with the further increase in the optical power intensity, the J_sc begins to show a slow growth trend, which is well consistent with the trend of J_sc inserted in Fig. 5(b). It can be explained that the CIPS is a room-temperature ferroelectric with a Curie temperature of 315 K. When the device is heated above 315 K, the ferroelectric CIPS will transform into the paraelectric phase with restored high symmetry and strictly prohibits the bulk photovoltaic effect.²⁸ The decrease of J_sc near the Curie temperature indicates that the observed bulk photovoltaic effect is associated with the ferroelectricity of CIPS. The J_sc at 310 K is about 0.024 mA/cm², which is more than an order of magnitude smaller than caused by light, so the bulk photovoltaic effect in CIPS is temperature and light power dependent, but the effect of temperature is very small relative to light.

What is interesting is that higher temperature and higher voltage bias have a negative effect on J_ds in Fig. 5(a). This effect is more pronounced in the negative voltage region in Fig. 5(a), where the current decreases significantly. This phenomenon is known as negative differential resistance effect, which is characterized by a decrease in current with increasing voltage.³⁰ When a negative voltage is applied, there is an increase in charge injection at the CIPS/Au interface, leading to an increase in current. However, this increase in current is subsequently suppressed by the trapping of electrons at the CIPS/MoS₂ interface. As a result, the current does not continue to increase at higher negative voltage regions.^31–33 Moreover, at higher temperatures, the polarization of the CIPS decreases.^28,34 This decrease in polarization reduces the number of trapped electrons at the CIPS/MoS₂ interface, weakening the negative differential resistance effect. The decreasing polarization also affects the CIPS/Au interface, reducing charge injection and causing a decrease in current as the temperature increases. Furthermore, the decreasing polarization as the temperature also suppresses the rectification characteristics, as shown in Fig. 5(a). The combination of the CIPS/Au and CIPS/MoS₂ interfaces, along with the polarization in the CIPS, results in the fact that the J_ds–V_ds curves are negatively affected by higher temperatures and higher voltage biases in the negative voltage region.

We summarized the J_sc generated via bulk photovoltaic effect from different materials in Fig. 6(a).^{9,20,28,35–39} Although the performance of the device is not the best in similar area, J_sc of 1.23 mA/cm² is several orders of magnitude or comparable to that of other materials. The detection of weak light can be realized by adding the MoS₂ conductive layer. For 3D materials, most of the ferroelectrics are oxide insulators with perovskite structures, where the large gap width fundamentally limits the photocurrent density, which highlights the potential advantages of CIPS/MoS₂ heterojunctions for bulk photovoltaic effect. To compare the performance of the CIPS/MoS₂ heterojunction at different wavelengths, we measured the current–time curves under 405, 808, and 1064 nm lasers without external bias, as shown in Figs. 6(b)–6(d). Under 405 nm laser, this zero-bias photocurrent is regarded as a characteristic feature of the bulk photovoltaic effect, and self-powered detection is realized. The current generated by the 808 nm laser is an order of magnitude smaller, which is due to the bandgap limitation of the CIPS. MoS₂ is an indirect bandgap semiconductor material, bandgap of 1.29 eV. Layer as the reduction of the number of less, due to quantum confinement effect, indirect bandgap increases gradually, to the single layer became a direct bandgap of 1.9 eV. Therefore, the 1064 nm laser exceeds the bandgap of MoS₂, and the current is only 1.5 pA at maximum optical power intensity, which can be attributed to the effect of heat generated by the laser.

Finally, we evaluated the photodetection performance of the heterojunction. Figure 7(a) shows the photocurrent of CIPS/MoS₂ heterojunction at 405 nm. The photocurrent (I_ph) increases with the rise of optical power intensity and the maximum value is up to 113 nA. Here, I_ph is defined as the difference between the current under light illumination and that in the dark at the bias of 3 V. It is calculated using the same experiment data, as shown in Fig. 3(a). The photoresponsivity (R), external quantum efficiency (EQE), and D^* are calculated to further evaluate the performance. R is one of the core parameters of a photodetector, representing the ability of a photodetector to convert the optical signal into an electrical signal at a specific optical power intensity and wavelength.^40, R is calculated as R = I_ph/PS, where P is optical power intensity irradiated on the heterojunction, and S is the effective channel area of the device, which is about 735 μm². EQE represents the ratio of the number of photogenerated electrons produced by the photodetector to the number of photons irradiated on the photodetector per unit time. EQE can be calculated as EQE = hcR/eλ, where c is the velocity of light, h is Planck's constant, e is the elemental charge (1.602 × 10⁻¹⁹ C), and λ is the wavelength of the light. D^* indicates the signal-to-noise ratio of the photodetector, which can evaluate the ability of the photodetector to detect weak signals.^40,41 D^* can be calculated by D^* = /(2eI_dark/S)^1/2, where I_dark is the current in the dark. Figures 7(b)–7(d) depict the R, EQE, and D^* as a function of optical power intensity, respectively. The tendency of them is decline first, then slightly increased and decreased again as the laser intensity increases. When the power density is 4.71 mW/cm², the R, EQE, and D^* of the heterojunction reach the maximum values, which are 41 mA/W, 0.13%, and 1.89 × 10⁹ Jones, respectively. Table I shows a comparison of the core performance parameters of our work with other materials recently reported, including measured wavelengths, I_ph, R, and D^*. It can be seen from the table that the CuInP₂S₆/MoS₂ heterojunction as-fabricated possess comparable I_ph, R, and D^*. Similarly, the performance of our devices is not the best in the similar area, but we provide an idea to improve device performances of and highlight the potential for high-performance optoelectronic devices based on ferroelectric vdW CuInP₂S₆/MoS₂ heterojunction.

In summary, we report an efficient optoelectronic device based on CIPS/MoS₂ heterojunction, and the photoelectric effect in the MoS₂ layer contributes to the photoconductivity effect of the device, while the ferroelectric material CIPS contributes to the bulk photovoltaic effect associated with the room-temperature polar ordering. The J_sc of the device is reaching about 1.23 mA/cm² when the optical power intensity is 0.35 W/cm², J_sc is about 0.024 mA/cm² at 310 K that is an order of magnitude smaller than that caused by light, so photoconductivity effect is main and the pyroelectric effect can be excluded. The corresponding R, EQE, and D^* are 41 mA/W, 0.13%, and 1.89 × 10⁹ Jones at the optical power intensity for 4.71 mW/cm², respectively. Our findings reveal that the CIPS exhibit significantly enhanced photovoltaic performance and offer a significant potential for developing the photodetector and then greatly enrich the functionalities and potential applications of the self-powered ferroelectric modulated photodetector.

This work was supported by the National Natural Science Foundation of China (No. 12175191), the Science and Technology Innovation Program of Hunan Province, China (Grant No. 2023RC3134), the Natural Science Foundation of Hunan Province, China (Nos. 2022JJ30566), and the Research Foundation of Education Bureau of Hunan Province, China (Grant No. 22A0134).